account-scoring
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the
@cargo-ai/clipackage, which is an official deployment tool provided by the vendor. - [PROMPT_INJECTION]: The
account-scoreragent inagents/scorer.tsexhibits an indirect prompt injection surface because it processes untrusted data from external business database lookups. - Ingestion points: Business records and firmographics are ingested via the
cargoDb.actions.matchBusinessandenrichBusinessFirmographicstools inagents/scorer.tsandplays/score-accounts.ts. - Boundary markers: The system prompt for the agent does not define clear boundaries or provide instructions to disregard potentially malicious instructions embedded within retrieved business data.
- Capability inventory: The agent can perform business database lookups, and the resulting scores and rationales are written to the user's CRM via the HubSpot connector.
- Sanitization: The skill relies on the agent's output schema to structure the response but does not perform explicit sanitization of the input business data before it is processed by the LLM.
Audit Metadata